Modelled a light, basically a cylinder with a bit of poly modelling and edge chamfering to make it resemble the light fitting.
1st render has a multi/sub material on it with the bottom part of the mesh as the vraylightmtl and the top part a plasticy type material. (its all a single mesh). Vraylightmtl is set to 20
2nd render has the same multi/sub material but the vraylightmtl has been replaced by a white vray material. And i have added a sphere vraylight just below the light fitting and have scaled the vraylight down on the Z axis so that it resembles a thin disc. The vray light has a multiplyer of 15 and 20subdivs also tested with 100subs to remove grain.
Question is, is how do i make the vraylightmtl render more “accurate” and “sharp”.. more like the vraylight one. How does one control the subdivs of the vraylightmtl?? via the mesh? more faces = more subdivision for the vraylightmtl or something?
Or am I just asking too much of the vraylightmtl?
Render settings.
Adaptive QMC AA 2-5
IR: -3,-1 , 0.2,0.2,0.2 . 50,30
LC: 1000subs, sample size 0.1, Filter = fixed 0.05
No other lights in the scene.
To get a smoother image with the VRayLightMtl, you need to raise the IR Hsph.value. Yours is 50, try 100 - will definately have different look and rendertime as well.
Thanks Nikki, I rendered it at 100 and it doubled the render time, and its still not acceptable. Especially near the shelves, you can see its nowhere near accurate as well as there is still splotches on the roof.
Plus the rendertime jump certainly isnt worth the slightly more real looking light cast, given I can simulate a very similar effect using my other way and the render time should be much much lower
Any other suggestions? other than keep cranking up the Hsph value, which i think is a little extreme @ 100 already.
It’s normal that vraylightmtl is slower, your making a true object light here, so quality depends completely on your GI settings, while with a vray light it is direct light.
Clr=0.2 is way too low. Going from 0.3 to 0.2 has a huge effect on rendertime, the relationship clr th vs rendertime is not at all linear, and especially below 0.3 rendertimes go crazy, there are few cases that such a low clr th is needed.
Here’s a graph I once made from a testscene. The scene contained both detailed and non-detailed areas to make the test a bit fair.
Will bump it up to 0.3 and see what happens, if the affect is big enough might move back to the vraylightmtl but i dont think it will knock enough rendertime off.
No indeed you will never get the same amount of detail with the same rendertime with your vraylightmtl. Why would you prefer using the mtl instead of the workaround?
Well was kinda going for a more accurate light simulation… but i guess they look pretty similar.
But yeah if I want resonable render times I gotta stick with the work around… which for the most part looks good enough. Cause I guess on a scene like this you are not really gonna notice it anyway… so meh… sod it.. work around it is
Ok, I think it will have a much bigger effect on rendertimes with the vraylightmtl, but you will probably see a quality loss.
Are you still using 100 subdivs on your vraylight? What is your QMC noise th? I usually prefer lower subdivs on the light and to clear out the noise, use a lower QMC noise th (like 0.002 instead of default 0.005).
